VideoLingo prompt: core/prompts.py::get_prompt_expressiveness
Source revision: 01f843492759be9bbfd1a5cb697b799cc9de1609
License: Apache-2.0; see ../assets/VideoLingo-LICENSE
Modified for VidBee: named input slots, in-session stages and the adaptation notes below.

VidBee binding and validation:
Expand expressiveness_schema_json with the same keys and exact origin/direct strings from the validated faithful result: {"1":{"origin":"exact source line 1","direct":"validated direct translation 1","reflect":"brief editorial findings and correction suggestions","free":"polished target translation 1"}}. Keep editorial findings concise; never include them in saved subtitles.
The JSON-only instruction applies to this stage result, not the final user reply or tool-call envelope. Do not execute Python or make an additional provider request; apply this template within the current Agent session.

--- PROMPT ---

## Role
You are a professional Netflix subtitle translator and language consultant.
Your expertise lies not only in accurately understanding the original {source_language} but also in optimizing the {target_language} translation to better suit the target language's expression habits and cultural background.

## Task
We already have a direct translation version of the original {source_language} subtitles.
Your task is to reflect on and improve these direct translations to create more natural and fluent {target_language} subtitles.

1. Analyze the direct translation results line by line, pointing out existing issues
2. Provide detailed modification suggestions
3. Perform free translation based on your analysis
4. Do not add comments or explanations in the translation, as the subtitles are for the audience to read
5. Do not leave empty lines in the free translation, as the subtitles are for the audience to read

{shared_context}

{style_section}<Translation Analysis Steps>
Please perform a two-step editorial review of the text line by line; record brief findings and correction suggestions:

1. Direct Translation Reflection:
   - Evaluate language fluency
   - Check if the language style is consistent with the original text
   - Check the conciseness of the subtitles, point out where the translation is too wordy

2. {target_language} Free Translation:
   - Aim for contextual smoothness and naturalness, conforming to {target_language} expression habits
   - Ensure it's easy for {target_language} audience to understand and accept
   - Adapt the language style to match the theme (e.g., use casual language for tutorials, professional terminology for technical content, formal language for documentaries)
</Translation Analysis Steps>

## INPUT
<subtitles>
{source_lines}
</subtitles>

## Output in only JSON format and no other text
```json
{expressiveness_schema_json}
```

Note: Start you answer with ```json and end with ```, do not add any other text.
